Questions on LMM #3 Flashcards
(12 cards)
True or false : Aliasing means that an infinite number of combination of parameters maximize the likelihoos, which can cause estimation problems
True
How do we solve the aliasing problem?
By adding some constraints
True or false : In general, analyses using LMMs are carried out under the assumption that missing data in clustered or longitudinal data sets are missing at random (MAR)
True
True or false : Under the MAR assumption, the ML and REML estimation are not valid
False : ML and REML are valid
Data is considered missing at random if..
The probability that it is missing is not a function of the data itself
True or false : Missing data is more prevalent for longitudinal studies, when subjects may drop out or not report values for other reasons at certain time
True
What is the other option to handle missing data for repeated measure or longitudinal studies?
Multivariate ANOVA
What are the 2 advantages LMM over Anova for handling missing data?
- ANOVA omits all observations for a subject if data is missing at some time points. LMM do not
- ANOVA requires observations for all subjects at each time point, whereas LMMs can handle different time points for different subjects
True or false : If most subjects have observations at one time point only, LMM should not be used. It would be difficult to estimate random factors for each subject
TRUE
In addition of the 2 advantages of LMM over ANOVA for missing data, can you tell me 3 more advantages of LMM over ANOVA?
- The covariance structures of the residuals is compound symmetry for ANOVA. LMM is more flexible
- LMM allows inclusion of time-varying covariates other than time itself. ANOVA doesn’t
- LMM allows random coefficients, ANOVA only allows random intercept
Tell me more of grand mean centering
- The overall mean of the covariate is substracted from it. Does not affect the fitted coefficient, but will affect the intercept
- Before centering, the intercept is the expected value of the response when the covariate is 0
- After centering, the intercept is the expected value of the response when the covariate equals its mean
Tell me more of group mean centering
- The mean of the covariate within its level-2 group is substracted from it. Does affect the coefficient and intercept
- Before centering, the coefficient represents the change in the response for each unit change of the covariate
- After centering, the coefficient represents the change in the response for each unit change of the covariate within the group
